MemoryCache empty just after adding an object - asp.net-mvc

I'm facing a strange problem with ASP.NET MemoryCaching in a MVC 3 ASP.NET application.
Each time an action is executed, I check if its LoginInfo are actually stored in the MemoryCache (code has been simplified, but core is as follow):
[NonAction]
protected override void OnAuthorization(AuthorizationContext filterContext) {
Boolean autorizzato = false;
LoginInfo me = CacheUtils.GetLoginData(User.Identity.Name);
if (me == null)
{
me = LoginData.UserLogin(User.Identity.Name);
CacheUtils.SetLoginInfo(User.Identity.Name, me);
}
// Test if the object is really in the memory cache
if (CacheUtils.GetLoginData(User.Identity.Name) == null) {
throw new Exception("IMPOSSIBLE");
}
}
The GetLoginInfo is:
public static LoginInfo GetLoginData(String Username)
{
LoginInfo local = null;
ObjectCache cache = MemoryCache.Default;
if (cache.Contains(Username.ToUpper()))
{
local = (LoginInfo)cache.Get(Username.ToUpper());
}
else
{
log.Warn("User " + Username + " not found in cache");
}
return local;
}
The SetLoginInfo is:
public static void SetLoginInfo (String Username, LoginInfo Info)
{
ObjectCache cache = MemoryCache.Default;
if ((Username != null) && (Info != null))
{
if (cache.Contains(Username.ToUpper()))
{
cache.Remove(Username.ToUpper());
}
cache.Add(Username.ToUpper(), Info, new CacheItemPolicy());
}
else
{
log.Error("NotFound...");
}
}
The code is pretty straightforward, but sometimes (totally randomly), just after adding the LoginInfo to the MemoryCache, this results empty, the just added Object is not present, therefore I got the Exception.
I'm testing this both on Cassini and IIS 7, it seems not related to AppPool reusability (enabled in IIS 7), I've tested with several Caching policies, but cannot make it work
What Am I missing/Failing ?
PS: forgive me for my bad english

Looking at the code for MemoryCache using a decomplier there is the following private function
private void OnUnhandledException(object sender, UnhandledExceptionEventArgs eventArgs)
{
if (!eventArgs.IsTerminating)
return;
this.Dispose();
}
There is an unhandled exception handler setup by every MemoryCache for the current domain Thread.GetDomain() so if there is ever any exception in your application that is not caught which may be common in a website it disposes the MemoryCache for ever and cannot be reused this is especially relevant for IIS apps as apposed to windows applications that just exit on unhanded exceptions.

The MemoryCache has limited size. For the Default instance, is't heuristic value (according to MSDN).
Have you tried to set Priority property on CacheItemPolicy instance to NotRemovable?
You can have race-condition because the Contains-Remove-Add sequence in SetLoginInfo is not atomic - try to use Set method instead.
Btw. you are working on web application so why not to use System.Web.Caching.Cache instead?

I believe you are running into a problem that Scott Hanselman identified as a .NET 4 bug. Please see here: MemoryCache Empty : Returns null after being set

Related

Implementing ASP.NET MVC error handling using Castle Windsor's Dynamic Proxies

I spent a long time trying to get the ASP.NET MVC [HandleError] attribute to work in my websites. It seemed like a good idea to go with the solution offered by the framework, but I just couldn't get it to do anything useful. Then I tried writing my own attribute (mainly so that I could step in to the code with the debugger), but although my code seemed to be doing all the right things, after it executed the framework took over and did mysterious things. Finally I tried the MVC Contrib's [Rescue] attribute, which was better but I still couldn't get it to do what I wanted.
One problem is that exceptions thrown in code embedded in aspx / ascx pages get wrapped in HttpException's and WebHttpException's.
Another problem for me was that the system is very opaque. I was essentially poking inputs in to a black box with some desired outputs in mind, but with no idea (other than the documentation, which doesn't seem very accurate / thorough) what the relationship was between them.
So, what to do?
I went for Dynamic Proxies in Castle Windsor, using the code below, which tries to handle Database errors, for which I have a specific Exception (AccessDBException).
The _alreadyAttemptedToShowErrorPage is to stop infinite recursion in the case where the error page throws an Exception.
The GetAccessDBException(...) method finds the relevant exception anywhere in the Exception stack, for the case when there are problems in aspx / ascx code.
The code requires that there is a BaseController class that all controllers derive from. This class is used to add a CreateErrorView(...) method (being as the standard View(...) method is protected)
public class AccessDBExceptionHandlingDynamicProxy : IInterceptor
{
private bool _alreadyAttemptedToShowErrorPage;
public AccessDBExceptionHandlingDynamicProxy()
{
_alreadyAttemptedToShowErrorPage = false;
}
public void Intercept(IInvocation invocation)
{
Contract.Requires(invocation.Proxy is BaseController);
try
{
invocation.Proceed();
}
catch (HttpException e)
{
if (_alreadyAttemptedToShowErrorPage == true) throw e;
_alreadyAttemptedToShowErrorPage = true;
var dbException = GetAccessDBException(e);
if (dbException != null)
{
var baseController = (invocation.Proxy as BaseController);
var view = baseController.CreateErrorView("AccessDBException", new AccessDBExceptionViewModel(dbException));
baseController.Response.Clear();
baseController.Response.StatusCode = (int) HttpStatusCode.InternalServerError;
view.ExecuteResult(baseController.ControllerContext);
baseController.Response.End();
}
else
{
throw e;
}
}
}
private static AccessDBException GetAccessDBException(HttpException e)
{
AccessDBException dbException = null;
Exception current = e;
while (dbException == null && current != null)
{
if (current is AccessDBException) dbException = (current as AccessDBException);
current = current.InnerException;
}
return dbException;
}
}

Session becomes null in different thread

I have a task that runs in a different thread and requires the session. I've done:
public GenerateDocList(LLStatistics.DocLists.DocList docs)
{
this.docs = docs;
context = HttpContext.Current;
}
and
public void StartTask()
{
//this code runs in a separate thread
HttpContext.Current = context;
/* rest of the code */
}
Now the thread has knowledge of the session and it works for a while but at some point in my loop HttpContext.Current.Session becomes null. Any ideas what can I do about this?
public static LLDAC.DAL.DBCTX LLDB
{
get
{
LLDAC.DAL.DBCTX currentUserDBContext = HttpContext.Current.Session["LLDBContext"] as LLDAC.DAL.DBCTX;
if (currentUserDBContext == null)
{
currentUserDBContext = new LLDAC.DAL.DBCTX();
HttpContext.Current.Session.Add("LLDBContext", currentUserDBContext);//this works only for a few loop iterations
}
return currentUserDBContext;
}
}
In general, this is a very fragile pattern for a multi-threaded operation. Long-running tasks (which I assume this is) are best suited to instance methods in a class rather than static methods such that the class can maintain any dependent objects. Also, since the session state is not thread safe and can span multiple requests you are getting into some very risky business by cashing your DB context in the session at all.
If you are convinced this is best done with static methods and stored in the session, you may be able to do something like this:
public static HttpSessionState MySession { get; set; }
public GenerateDocList(LLStatistics.DocLists.DocList docs)
{
this.docs = docs;
MySession = HttpContext.Current.Session;
}
Then:
public static LLDAC.DAL.DBCTX LLDB
{
get
{
LLDAC.DAL.DBCTX currentUserDBContext = MySession["LLDBContext"] as LLDAC.DAL.DBCTX;
if (currentUserDBContext == null)
{
currentUserDBContext = new LLDAC.DAL.DBCTX();
if (MySession == null)
{
thow new InvalidOperaionException("MySession is null");
}
MySession.Add("LLDBContext", currentUserDBContext);
}
return currentUserDBContext;
}
}
Note that you could still run into issues with the session since other threads could still modify the session.
A better solution would probably look something like this:
public class DocListGenerator : IDisposable
{
public LLDAC.DAL.DBCTX LLDB { get; private set; }
public DocListGenerator()
{
LLDB = new LLDAC.DAL.DBCTX();
}
public void GenerateList()
{
// Put loop here.
}
public void Dispose()
{
if (LLDB != null)
{
LLDB.Dispose();
}
}
}
Then your calling code looks like this:
public void StartTask()
{
using (DocListGenerator generator = new DocListGenerator()
{
generator.GenerateList();
}
}
If you really want to cache something, you could cache your instance like this:
HttpContext.Current.Sesssion.Add("ListGenerator", generator);
However, I still don't think that is a particularly good idea since your context could still be disposed or otherwise altered by a different thread.
Using anything related to the HttpContext.Current on anything besides the main Request thread is generally going to get you into trouble in ASP.net.
The HttpContext is actually backed on a thread belonging to a Thread Pool and the thread may very well get reused on another request.
This is actually a common issue with using the new Async/Await keywords in ASP.net as well.
In order to help you, it would help to know why you're attempting this in the first place?
Is this a single server or a web farm with multiple load balanced servers?
Are you hosting it yourself, or is it the site hosted by a provider?
What is the SessionState implementation (SQL Server, State Server, In-Process, or something custom like MemCached, Redis, etc...)
What version of ASP .net?
Why are you starting a new thread instead of just doing the processing on the request thread?
If you really can't (or shouldn't) use session. Then you could use something like a correlation ID.
Guid correlationID = Guid.NewGuid();
HttpContext.Current.Session["DocListID"] = correlationID;
DocList.GoOffAndGenerateSomeStuffOnANewThread(correlationID);
... when process is done, store the results somewhere using the specified ID
// Serialize the result to SQL server, the file system, cache...
DocList.StoreResultsSomewhereUnderID();
... later on
DocList.CheckForResultsUnderID(HttpContext.Current.Session["DocListID"]);

Concurrent requests: session lost after removing session-id from one single request

In my ASP .NET MVC 2 - application, there are several controllers, that need the session state. However, one of my controllers in some cases runs very long and the client should be able to stop it.
Here is the long running controller:
[SessionExpireFilter]
[NoAsyncTimeout]
public void ComputeAsync(...) //needs the session
{
}
public ActionResult ComputeCompleted(...)
{
}
This is the controller to stop the request:
public ActionResult Stop()
{
...
}
Unfortunately, in ASP .NET MVC 2 concurrent requests are not possible for one and the same user, so my Stop-Request has to wait until the long running operation has completed. Therefore I have tried the trick described in this article and added the following handler to Global.asax.cs:
protected void Application_BeginRequest()
{
if (Request.Url.AbsoluteUri.Contains("Stop") && Request.Cookies["ASP.NET_SessionId"] != null)
{
var session_id = Request.Cookies["ASP.NET_SessionId"].Value;
Request.Cookies.Remove("ASP.NET_SessionId");
...
}
}
This simply removes the session-id from the Stop-Request. At the first glance this works well - the Stop-Request comes through and the operation is stopped. However, after that, it seems that the session of the user with the long running request has been killed.
I use my own SessionExpireFilter in order to recognize session timeouts:
public class SessionExpireFilterAttribute : ActionFilterAttribute
{
public override void OnActionExecuting(ActionExecutingContext filterContext)
{
HttpContext ctx = HttpContext.Current;
// check if session is supported
if (ctx.Session != null)
{
// check if a new session id was generated
if (ctx.Session.IsNewSession)
{
// If it says it is a new session, but an existing cookie exists, then it must
// have timed out
string sessionCookie = ctx.Request.Headers["Cookie"];
if ((null != sessionCookie) && (sessionCookie.IndexOf("ASP.NET_SessionId") >= 0))
{
filterContext.Result = new JsonResult() { Data = new { success = false, timeout = true }, JsonRequestBehavior = JsonRequestBehavior.AllowGet };
}
}
}
base.OnActionExecuting(filterContext);
}
}
ctx.Session.IsNewSession is always true after the Stop-Request has been called, but I don't know why. Does anyone know why the session is lost? Is there any mistake in the implementation of the Stop-Controller?
The session is lost because you removed the session cookie. I'm not sure why that seems illogical. Each new page request supplies the cookie to asp.net, and if there is no cookie it generates a new one.
One option you could use to use cookieless sessions, which will add a token to the querystring. All you need to do is generate a new session for each login, or similar.
But this is one of the reasons why session variables are discouraged. Can you change the code to use an in-page variable, or store the variable in a database?

Getting more "granularity" from the MVC Mini Profiler

Should this turn out to be useful for anyone, I'll gladly turn it into a community wiki thing.
I have some slow pages in an MVC3 app, and since little of the execution time seemed to happen in my code, I wanted to see if I could find out more about what took so long. Not that I succeeded, but I gained a little more wisdom along the way.
There is nothing here that isn't obvious to anyone with some MVC experience. Basically, I created my own ActionFilterAttribute that looks like this:
public class ProfilerAttribute : ActionFilterAttribute
{
IDisposable actionStep = null;
IDisposable resultStep = null;
public override void OnActionExecuting(ActionExecutingContext filterContext)
{
actionStep = MiniProfiler.Current.Step("OnActionExecuting " + ResultDescriptor(filterContext));
base.OnActionExecuting(filterContext);
}
public override void OnActionExecuted(ActionExecutedContext filterContext)
{
if (actionStep != null)
{
actionStep.Dispose();
actionStep = null;
}
base.OnActionExecuted(filterContext);
}
public override void OnResultExecuting(ResultExecutingContext filterContext)
{
resultStep = MiniProfiler.Current.Step("OnResultExecuting " + ResultDescriptor(filterContext));
base.OnResultExecuting(filterContext);
}
public override void OnResultExecuted(ResultExecutedContext filterContext)
{
if (resultStep != null)
{
resultStep.Dispose();
resultStep = null;
}
base.OnResultExecuted(filterContext);
}
private string ResultDescriptor(ActionExecutingContext filterContext)
{
return filterContext.ActionDescriptor.ControllerDescriptor.ControllerName + "." + filterContext.ActionDescriptor.ActionName;
}
private string ResultDescriptor(ResultExecutingContext filterContext)
{
var values = filterContext.RouteData.Values;
return String.Format("{0}.{1}", values["controller"], values["action"]);
}
This seems to work well, and in my case I have learned that most of the time is actually spent in the ResultExecuting part of life, not inside my actions.
However, I have some questions about this approach.
1) Is this a request-safe way of doing things? I am guessing no, since the actionfilter is created only once, in the RegisterGlobalFilters() method in Global.asax.cs. If two requests appear at once, actionStep and resultStep will be worthless. Is this true? If so, can someone who knows more than me contribute a clever way to handle this? Works for me during local machine profiling, but probably not so much deployed on a server with multiple people making requests at the same time.
2) Is there any way to get more insight into the result-executing process? Or should I just accept that rendering the view etc. takes the time it takes? In my own app I ensure that all database access is finished before my action method is over (using NHibernate Profiler in my case), and I like to keep my views slim and simple; Any kind of insight into what slows the rendering down could still be useful, though. I guess using the Mini Profiler in my model objects would show up here, if any slow code on my part was executed here.
3) The ResultDescriptor methods are probably evil and poisonous. They've worked for me in my tests, but would probably need to be replaced by something more robust. I just went with the first versions that gave me something halfway useful.
Any other comments to this would also be very welcome, even if they are "This is a bad idea, go die alone".
This looks like a cool idea. I believe that it's NOT a request safe way of doing things.
You could link it to HttpContext.Items like this
HttpContext.Items.Add("actionstep", actionStep);
HttpContext.Items.Add("resultstep", resultStep);
And then retrieve it in similar fashion
actionStep = HttpContext.Items["actionstep"];
resultStep = HttpContext.Items["resultstep"];
Obviously putting in your own checks for nulls and so forth.
The HttpContext is different for each user/request.
The thing to remember about HttpContext.Current.Session.SessionID which I sometimes forget it that it is the SessionId of the current HTTP request (i.e. it changes each time you hit F5 or otherwise make a new request). The other important thing to remember is that, whilst at any on time, all HttpContext.Current.Session.SessionID values are necessarily unique (i.e. one for each user, or request), they can be reused, so dno't think of them as GUIDs which are only used once each.
There is already an action filter attribute in the MiniProfiler assembly that does the profiling for the actions. It is in the StackExchange.Profiling.MVCHelpers namespace and it's called ProfilingActionFilter. You can extend it to also profile your views.
It uses the same approach as described by #Dommer but instead of storing the IDisposable directly, it stores a Stack in the HttpContext.Current.Items. You can do the same for the views.
Here is the code for the action profiling:
/// <summary>
/// This filter can be applied globally to hook up automatic action profiling
///
/// </summary>
public class ProfilingActionFilter : ActionFilterAttribute
{
private const string stackKey = "ProfilingActionFilterStack";
/// <summary>
/// Happens before the action starts running
///
/// </summary>
public override void OnActionExecuting(ActionExecutingContext filterContext)
{
if (MiniProfiler.Current != null)
{
Stack<IDisposable> stack = HttpContext.Current.Items[(object) "ProfilingActionFilterStack"] as Stack<IDisposable>;
if (stack == null)
{
stack = new Stack<IDisposable>();
HttpContext.Current.Items[(object) "ProfilingActionFilterStack"] = (object) stack;
}
MiniProfiler current = MiniProfiler.Current;
if (current != null)
{
RouteValueDictionary dataTokens = filterContext.RouteData.DataTokens;
string str1 = !dataTokens.ContainsKey("area") || string.IsNullOrEmpty(dataTokens["area"].ToString()) ? "" : (string) dataTokens["area"] + (object) ".";
string str2 = Enumerable.Last<string>((IEnumerable<string>) filterContext.Controller.ToString().Split(new char[1] { '.' })) + ".";
string actionName = filterContext.ActionDescriptor.ActionName;
stack.Push(MiniProfilerExtensions.Step(current, "Controller: " + str1 + str2 + actionName, ProfileLevel.Info));
}
}
base.OnActionExecuting(filterContext);
}
/// <summary>
/// Happens after the action executes
///
/// </summary>
public override void OnActionExecuted(ActionExecutedContext filterContext)
{
base.OnActionExecuted(filterContext);
Stack<IDisposable> stack = HttpContext.Current.Items[(object) "ProfilingActionFilterStack"] as Stack<IDisposable>;
if (stack == null || stack.Count <= 0) return;
stack.Pop().Dispose();
}
}
Hope this help.
You can just wrap ExecuteCore method on Controller. :)

ELMAH - Using custom error pages to collecting user feedback

I'm looking at using ELMAH for the first time but have a requirement that needs to be met that I'm not sure how to go about achieving...
Basically, I am going to configure ELMAH to work under asp.net MVC and get it to log errors to the database when they occur. On top of this I be using customErrors to direct the user to a friendly message page when an error occurs. Fairly standard stuff...
The requirement is that on this custom error page I have a form which enables to user to provide extra information if they wish. Now the problem arises due to the fact that at this point the error is already logged and I need to associate the loged error with the users feedback.
Normally, if I was using my own custom implementation, after I log the error I would pass through the ID of the error to the custom error page so that an association can be made. But because of the way that ELMAH works, I don't think the same is quite possible.
Hence I was wondering how people thought that one might go about doing this....
Cheers
UPDATE:
My solution to the problem is as follows:
public class UserCurrentConextUsingWebContext : IUserCurrentConext
{
private const string _StoredExceptionName = "System.StoredException.";
private const string _StoredExceptionIdName = "System.StoredExceptionId.";
public virtual string UniqueAddress
{
get { return HttpContext.Current.Request.UserHostAddress; }
}
public Exception StoredException
{
get { return HttpContext.Current.Application[_StoredExceptionName + this.UniqueAddress] as Exception; }
set { HttpContext.Current.Application[_StoredExceptionName + this.UniqueAddress] = value; }
}
public string StoredExceptionId
{
get { return HttpContext.Current.Application[_StoredExceptionIdName + this.UniqueAddress] as string; }
set { HttpContext.Current.Application[_StoredExceptionIdName + this.UniqueAddress] = value; }
}
}
Then when the error occurs, I have something like this in my Global.asax:
public void ErrorLog_Logged(object sender, ErrorLoggedEventArgs args)
{
var item = new UserCurrentConextUsingWebContext();
item.StoredException = args.Entry.Error.Exception;
item.StoredExceptionId = args.Entry.Id;
}
Then where ever you are later you can pull out the details by
var item = new UserCurrentConextUsingWebContext();
var error = item.StoredException;
var errorId = item.StoredExceptionId;
item.StoredException = null;
item.StoredExceptionId = null;
Note this isn't 100% perfect as its possible for the same IP to have multiple requests to have errors at the same time. But the likely hood of that happening is remote. And this solution is independent of the session, which in our case is important, also some errors can cause sessions to be terminated, etc. Hence why this approach has worked nicely for us.
The ErrorLogModule in ELMAH (version 1.1 as of this writing) provides a Logged event that you can handle in Global.asax and which you can use to communicate details, say via HttpContext.Items collection, to your custom error page. If you registered the ErrorLogModule under the name ErrorLog in web.config then your event handler in Global.asax will look like this:
void ErrorLog_Logged(object sender, ErrorLoggedEventArgs args)
{
var id = args.Entry.Id
// ...
}

Resources